S.

Boyd EE102

Lecture 5
Rational functions and partial fraction
expansion

• (review of) polynomials

• rational functions
• pole-zero plots
• partial fraction expansion
• repeated poles
• nonproper rational functions

5–1
 Polynomials and roots

polynomials
a(s) = a0 + a1s + · · · + ansn

• a is a polynomial in the variable s

• ai are the coefficients of a (usually real, but occasionally complex)
• n is the degree of a (assuming an 6= 0)

roots (or zeros) of a polynomial a: λ ∈ C that satisfy

a(λ) = 0

examples
• a(s) = 3 has no roots
3
√ √
• a(s) = s − 1 has three roots: 1, (−1 + j 3)/2, (−1 − j 3)/2

Rational functions and partial fraction expansion 5–2

factoring out roots of a polynomial

if a has a root at s = λ we can factor out s − λ:

• dividing a by s − λ yields a polynomial:

a(s)
b(s) =
s−λ

is a polynomial (of degree one less than the degree of a)

• we can express a as
a(s) = (s − λ)b(s)
for some polynomial b

example: s3 − 1 has a root at s = 1

s3 − 1 = (s − 1)(s2 + s + 1)

Rational functions and partial fraction expansion 5–3

the multiplicity of a root λ is the number of factors s − λ we can factor
out, i.e., the largest k such that

a(s)
(s − λ)k

is a polynomial

example:
a(s) = s3 + s2 − s − 1

• a has a zero at s = −1
a(s) s3 + s 2 − s − 1
• = = s2 − 1 also has a zero at s = −1
s+1 s+1
a(s) s3 + s 2 − s − 1
• = = s − 1 does not have a zero at s = −1
(s + 1)2 (s + 1)2

so the multiplicity of the zero at s = −1 is 2

Rational functions and partial fraction expansion 5–4

 Fundamental theorem of algebra

a polynomial of degree n has exactly n roots, counting multiplicities

this means we can write a in factored form

a(s) = ansn + · · · + a0 = an(s − λ1) · · · (s − λn)

where λ1, . . . , λn are the n roots of a

example: s3 + s2 − s − 1 = (s + 1)2(s − 1)

the relation between the coefficients ai and the λi is complicated in

general, but
Yn Xn
a0 = an (−λi), an−1 = −an λi
i=1 i=1
are two identities that are worth remembering

Rational functions and partial fraction expansion 5–5

 Conjugate symmetry

if the coefficients a0, . . . , an are real, and λ ∈ C is a root, i.e.,

a(λ) = a0 + a1λ + · · · + anλn = 0

then we have
n
a(λ) = a0 + a1λ + · · · + anλ = (a0 + a1λ + · · · + anλn) = a(λ) = 0

in other words: λ is also a root

• if λ is real this isn’t interesting

• if λ is complex, it gives us another root for free
• complex roots come in complex conjugate pairs

Rational functions and partial fraction expansion 5–6

example:
PSfrag replacements =
λ1
λ4
λ3 λ6
<
λ5
λ2

λ3 and λ6 are real; λ1, λ2 are a complex conjugate pair; λ4, λ5 are a
complex conjugate pair
if a has real coefficients, we can factor it as
à r !à m !
Y Y
a(s) = an (s − λi) (s − λi)(s − λi)
i=1 i=r+1

where λ1, . . . , λr are the real roots; λr+1, λr+1, . . . , λm, λm are the
complex roots

Rational functions and partial fraction expansion 5–7

 Real factored form

(s − λ)(s − λ) = s2 − 2(<λ) s + |λ|2

is a quadratic with real coefficients

real factored form of a polynomial a:

à r
!Ã m
!
Y Y
a(s) = an (s − λi) (s2 + αis + βi)
i=1 i=r+1

• λ1, . . . , λr are the real roots

• αi, βi are real and satisfy αi2 < 4βi

any polynomial with real coefficients can be factored into a product of

• degree one polynomials with real coefficients
• quadratic polynomials with real coefficients

Rational functions and partial fraction expansion 5–8

example: s3 − 1 has roots
√ √
−1 + j 3 −1 − j 3
s = 1, s= , s=
2 2

• complex factored form

³ √ ´³ √ ´
s3 − 1 = (s − 1) s + (1 + j 3)/2 s + (1 − j 3)/2

• real factored form

s3 − 1 = (s − 1)(s2 + s + 1)

Rational functions and partial fraction expansion 5–9

 Rational functions

a rational function has the form

b(s) b0 + b 1 s + · · · + b m s m
F (s) = = n
,
a(s) a0 + a 1 s + · · · + a n s

i.e., a ratio of two polynomials (where a is not the zero polynomial)

• b is called the numerator polynomial

• a is called the denominator polynomial

examples of rational functions:

1 2 1 s s3 + 3s + 3
, s + 3, + = 3
s+1 s + 1 2s + 3 2s + 3s2 + 2s + 3
2

Rational functions and partial fraction expansion 5–10

 b(s)
rational function F (s) =
a(s)

polynomials b and a are not uniquely determined, e.g.,

1 3 s2 + 3
= =
s + 1 3s + 3 (s + 1)(s2 + 3)
√
(except at s = ±j 3. . . )

rational functions are closed under addition, subtraction, multiplication,

division (except by the rational function 0)

Rational functions and partial fraction expansion 5–11

 Poles & zeros

b(s) b0 + b 1 s + · · · + b m s m
F (s) = = n
,
a(s) a0 + a 1 s + · · · + a n s

assume b and a have no common factors (cancel them out if they do . . . )

• the m roots of b are called the zeros of F ; λ is a zero of F if F (λ) = 0

• the n roots of a are called the poles of F ; λ is a pole of F if
lims→λ |F (s)| = ∞

the multiplicity of a zero (or pole) λ of F is the multiplicity of the root λ

of b (or a)
6s + 12
example: 2 has one zero at s = −2, two poles at s = −1
s + 2s + 1

Rational functions and partial fraction expansion 5–12

factored or pole-zero form of F :

b0 + b 1 s + · · · + b m s m (s − z1) · · · (s − zm)
F (s) = =k
a0 + a 1 s + · · · + a n sn (s − p1) · · · (s − pn)

where

• k = bm/an
• z1, . . . , zm are the zeros of F (i.e., roots of b)
• p1, . . . , pn are the poles of F (i.e., roots of a)

(assuming the coefficients of a and b are real) complex poles or zeros come
in complex conjugate pairs

can also have real factored form . . .

Rational functions and partial fraction expansion 5–13

 Pole-zero plots
poles & zeros of a rational functions are often shown in a pole-zero plot
=

j
PSfrag replacements 1
<

(× denotes a pole; ◦ denotes a zero)

this example is for
(s + 1.5)(s + 1 + 2j)(s + 1 − 2j)
F (s) = k
(s + 2.5)(s − 2)(s − 1 − j)(s − 1 + j)
(s + 1.5)(s2 + 2s + 5)
= k
(s + 2.5)(s − 2)(s2 − 2s + 2)

(the plot doesn’t tell us k)

Rational functions and partial fraction expansion 5–14

 Partial fraction expansion

b(s) b0 + b 1 s + · · · + b m s m
F (s) = =
a(s) a0 + a 1 s + · · · + a n sn

let’s assume (for now)

• no poles are repeated, i.e., all roots of a have multiplicity one
• m<n

then we can write F in the form

r1 rn
F (s) = + ··· +
s − λ1 s − λn
called partial fraction expansion of F
• λ1, . . . , λn are the poles of F
• the numbers r1, . . . , rn are called the residues
• when λk = λl, rk = rl

Rational functions and partial fraction expansion 5–15

example:

s2 − 2 −1 1 1
= + +
s3 + 3s2 + 2s s s+1 s+2

let’s check:
−1 1 1 −1(s + 1)(s + 2) + s(s + 2) + s(s + 1)
+ + =
s s+1 s+2 s(s + 1)(s + 2)

in partial fraction form, inverse Laplace transform is easy:

µ ¶
r1 rn
L−1(F ) = L−1 + ··· +
s − λ1 s − λn
= r 1 eλ1 t + · · · + r n eλn t

(this is real since whenever the poles are conjugates, the corresponding
residues are also)

Rational functions and partial fraction expansion 5–16

 Finding the partial fraction expansion

two steps:

• find poles λ1, . . . , λn (i.e., factor a(s))

• find residues r1, . . . , rn (several methods)

method 1: solve linear equations

we’ll illustrate for m = 2, n = 3

b0 + b1s + b2s2 r1 r2 r3
= + +
(s − λ1)(s − λ2)(s − λ3) s − λ1 s − λ2 s − λ3

Rational functions and partial fraction expansion 5–17

clear denominators:

b0 + b1s + b2s2 = r1(s − λ2)(s − λ3) + r2(s − λ1)(s − λ3) + r3(s − λ1)(s − λ2)

equate coefficients:

• coefficient of s0:

b0 = (λ2λ3)r1 + (λ1λ3)r2 + (λ1λ2)r3

• coefficient of s1:

b1 = (−λ2 − λ3)r1 + (−λ1 − λ3)r2 + (−λ1 − λ2)r3

• coefficient of s2:
b2 = r 1 + r 2 + r 3

now solve for r1, r2, r3 (three equations in three variables)

Rational functions and partial fraction expansion 5–18

method 2: to get r1, multiply both sides by s − λ1 to get

(s − λ1)(b0 + b1s + b2s2) r2(s − λ1) r3(s − λ1)

= r1 + +
(s − λ1)(s − λ2)(s − λ3) s − λ2 s − λ3

cancel s − λ1 term on left and set s = λ1:

b0 + b1λ1 + b2λ21
= r1
(λ1 − λ2)(λ1 − λ3)

an explicit formula for r1! (can get r2, r3 the same way)
in the general case we have the formula

rk = (s − λk )F (s)|s=λk

which means:
• multiply F by s − λk
• then cancel s − λk from numerator and denominator
• then evaluate at s = λk to get rk

Rational functions and partial fraction expansion 5–19

example:
s2 − 2 r1 r2 r3
= + +
s(s + 1)(s + 2) s s+1 s+2

• residue r1:
µ ¶¯ 2
¯
r2 s r3s ¯¯ s −2 ¯
r1 = r1 + + = ¯ = −1
s+1 s+2 ¯ s=0 (s + 1)(s + 2) ¯ s=0

• residue r2:
µ ¶¯ 2
¯
r1(s + 1) r3(s + 1) ¯
¯ s − 2 ¯¯
r2 = + r2 + = =1
s s+2 ¯
s=−1 s(s + 2) s=−1
¯

• residue r3:
µ ¶¯ 2
¯
r1(s + 2) r2(s + 2) ¯ s − 2 ¯¯
r3 = + + r3 ¯
¯ = =1
s s+1 s=−2 s(s + 1) s=−2
¯

so we have:
s2 − 2 −1 1 1
= + +
s(s + 1)(s + 2) s s+1 s+2

Rational functions and partial fraction expansion 5–20

method 3: another explicit and useful expression for rk is:

b(λk )
rk =
a0(λk )

to see this, note that

(s − λk )b(s) b(s) + b0(s)(s − λk ) b(λk )

rk = lim = lim 0
= 0
s→λk a(s) s→λ k a (s) a (λk )

where we used l’Hôpital’s rule in second line

example (previous page):

s2 − 2 s2 − 2
=
s(s + 1)(s + 2) s3 + 2s2 + 2s

hence, ¯
2
s − 2 ¯¯
r1 = 2 = −1
3s + 4s + 2 ¯ s=0

Rational functions and partial fraction expansion 5–21

 Example

let’s solve
v 000 − v = 0, v(0) = 1, v 0(0) = v 00(0) = 0

1. take Laplace transform:

s3V (s) − s2 −V (s) = 0

| {z }
L(v 000 )

2. solve for V to get

s2
V (s) = 3
s −1

3. the poles of V are the cuberoots of 1, i.e., ej2πk/3, k = 0, 1, 2

3
³ √ ´³ √ ´
s − 1 = (s − 1) s + 1/2 + j 3/2 s + 1/2 − j 3/2

Rational functions and partial fraction expansion 5–22

4. now convert V to partial fraction form

r1 r2 r2
V (s) = + √ + √
s−1 s+ +j
1 3
s + 1
− j 3
2 2 2 2

to find residues we’ll use

√
b(1) 1 b(−1/2 − j 3/2) 1
r1 = 0
= , r2 = √ =
a (1) 3 a0(−1/2 − j 3/2) 3

so partial fraction form is

1 1 1
3 3 3
V (s) = + √ + √
s−1 1
s+ +j 3 1
s+ −j 3
2 2 2 2

(check this by just multiplying out . . . )

Rational functions and partial fraction expansion 5–23

5. take inverse Laplace transform to get v:
√ √
1 t 1 (− 1 −j 3 1 1 3
v(t) = e + e 2 2 )t + e(− 2 +j 2 )t
3 3 3
√
1 t 2 −t 3
= e + e cos
2 t
3 3 2

6. check that v 000 − v = 0, v(0) = 1, v 0(0) = v 00(0) = 0

Rational functions and partial fraction expansion 5–24

 Repeated poles
now suppose
b(s)
F (s) =
(s − λ1)k1 · · · (s − λl)kl

• the poles λi are distinct (λi 6= λj for i 6= j) and have multiplicity ki

• degree of b less than degree of a

partial fraction expansion has the form

r1,k1 r1,k1−1 r1,1
F (s) = + + ··· +
(s − λ1)k1 (s − λ1)k1−1 s − λ1
r2,k2 r2,k2−1 r2,1
+ + + · · · +
(s − λ2)k2 (s − λ2)k2−1 s − λ2
rl,kl rl,kl−1 rl,1
+ ··· + + + · · · +
(s − λl)kl (s − λl)kl−1 s − λl
n residues, just as before; terms involve higher powers of 1/(s − λ)

Rational functions and partial fraction expansion 5–25

 1
example: F (s) = 2 has expansion
s (s + 1)

r1 r2 r3
F (s) = + +
s2 s s+1

inverse Laplace transform of partial fraction form is easy since

µ ¶
r r
L −1
= tk−1eλt
(s − λ)k (k − 1)!

same types of tricks work to find the ri,j ’s

• solve linear equations (method 1)

• can find the residues for nonrepeated poles as before

Rational functions and partial fraction expansion 5–26

example:
1 r1 r2 r3
= + +
s2(s + 1) s2 s s+1

we get (as before)

r3 = (s + 1)F (s)|s=−1 = 1

now clear denominators to get

r1(s + 1) + r2s(s + 1) + s2 = 1
(1 + r2)s2 + (r1 + r2)s + 1 = 1

which yields r2 = −1, r1 = 1, so

1 1 1
F (s) = − +
s2 s s + 1

Rational functions and partial fraction expansion 5–27

extension of method 2: to get ri,ki ,

• multiply on both sides by (s − λi)ki

• evaluate at s = λi

gives ¯
ki ¯
F (s)(s − λi) s=λi
= ri,ki

to get other r’s, we have extension

j
¯
1 d ¡ ¢
ki ¯
¯
F (s)(s − λi) ¯ = ri,ki−j
j! dsj s=λi

usually the ki’s are small (e.g., 1 or 2), so fortunately this doesn’t come up
too often

Rational functions and partial fraction expansion 5–28

example (ctd.):
r1 r2 r3
F (s) = + +
s2 s s+1

• multiply by s2:

2 1 r3 s 2
s F (s) = = r1 + r2 s +
s+1 s+1

• evaluate at s = 0 to get r1 = 1

• differentiate with respect to s:

µ 2
¶
1 d r3 s
− = r 2 +
(s + 1)2 ds s+1

• evaluate at s = 0 to get r2 = −1

(same as what we got above)

Rational functions and partial fraction expansion 5–29

 Nonproper rational functions

b(s) b0 + b 1 s + · · · + b m s m
F (s) = = n
,
a(s) a0 + a 1 s + · · · + a n s

is called proper if m ≤ n, strictly proper if m < n, nonproper if m > n

partial fraction expansion requires strictly proper F ; to find L−1(F ) for
other cases, divide a into b:

F (s) = b(s)/a(s) = c(s) + d(s)/a(s)

where
c(s) = c0 + · · · + cm−nsm−n, d = d 0 + · · · + d k sk , k<n

then
L−1(F ) = c0δ + · · · + cm−nδ (m−n) + L−1(d/a)

d/a is strictly proper, hence has partial fraction form

Rational functions and partial fraction expansion 5–30

example
5s + 3
F (s) =
s+1
is proper, but not strictly proper

5(s + 1) − 5 + 3 2
F (s) = =5− ,
s+1 s+1

so
L−1(F ) = 5δ(t) − 2e−t

in general,
• F strictly proper ⇐⇒ f has no impulses at t = 0
• F proper, not strictly proper ⇐⇒ f has an impulse at t = 0
• F nonproper ⇐⇒ f has higher-order impulses at t = 0
• m − n determines order of impulse at t = 0

Rational functions and partial fraction expansion 5–31

 Example

s4 + s3 − 2s2 + 1
F (s) =
s3 + 2s2 + s

1. write as a sum of a polynomial and a strictly proper rational function:

s(s3 + 2s2 + s) − s(2s2 + s) + s3 − 2s2 + 1

F (s) =
s3 + 2s2 + s
−s3 − 3s2 + 1
= s+ 3
s + 2s2 + s
−(s3 + 2s2 + s) + (2s2 + s) − 3s2 + 1
= s+
s3 + 2s2 + s
−s2 + s + 1
= s−1+ 3
s + 2s2 + s
−s2 + s + 1
= s−1+
s(s + 1)2

Rational functions and partial fraction expansion 5–32

2. partial fraction expansion

−s2 + s + 1 r1 r2 r3
= + +
s(s + 1)2 s s + 1 (s + 1)2

• determine r1: ¯
2
−s + s + 1 ¯¯
r1 = =1
(s + 1)2 ¯ s=0
• determine r3: ¯
2
−s + s + 1 ¯¯
r3 = =1
s ¯
s=−1
• determine r2:
µ 2
¶¯ 2
¯
d −s + s + 1 ¯¯ −s − 1 ¯¯
r2 = = = −2
ds s ¯
s=−1 s2 ¯ s=−1

(alternatively, just plug in some value of s other than s = 0 or s = −1:

¯
2
−s + s + 1 ¯¯ 1 r2 r3 r2 1
¯ = = r 1 + + = 1 + + =⇒ r2 = −2)
s(s + 1)2 ¯ 4 2 4 2 4
s=1

Rational functions and partial fraction expansion 5–33

3. inverse Laplace transform
µ ¶
1 2 1
L−1(F (s)) = L−1 s−1+ − +
s s + 1 (s + 1)2

= δ 0(t) − δ(t) + 1 − 2e−t + te−t

Rational functions and partial fraction expansion 5–34